Conventional omnidirectional cameras (also known as virtual reality cameras, spherical cameras, panorama cameras, immersive video cameras, or 360 cameras) present design challenges. A single lens cannot capture an entire spherical field of view. The purpose of omnidirectional cameras is to capture video in all directions surrounding the camera (i.e., 360 degrees in each axis). The video captured represents a complete view of a scene surrounding the person watching the video. A user typically uses a head-mounted display or an interactive video player to view the captured video on playback. The video orientation can be changed in any direction during playback.
Conventional solutions previously considered include placing a convex mirror in front of the lens to capture more of the scene. However, using a convex mirror in front of the lens only provides a 360 degree horizontal coverage while losing the top and bottom of a spherical field of view. Conventional solutions previously considered also include capturing images from multiple lenses into several video signals. However, using several video signals creates several issues such as issues with processing all the video signals using a single processor, producing multiple video files using a single processor, providing dedicated processors for each video signal, and managing multiple video files from various processors and video signals.
Several conventional approaches have been attempted by commercial products. Most conventional approaches fall into one of several categories. In one conventional approach, each lens is mounted on a sensor, which is connected to a dedicated processor. Each processor records one video file onto a storage medium. After recording is done, all files are transferred out of the storage medium to a computer or cloud. Stitching the multiple files to create a spherical field of view is only done as a post-production process.
In another conventional approach, each lens is mounted on a sensor. All of the sensors in the system are connected to a single processor. The processor records multiple video files onto a storage medium. After recording is done, all files are transferred out of the storage medium to a computer or cloud. Stitching the multiple files to create a spherical field of view is only done as a post-production process.
It would be desirable to implement an omnidirectional camera with multiple processors and/or multiple sensors connected to each processor.